Beruflich Dokumente
Kultur Dokumente
by
Etienne A. Alderlieste
April 2011
Version: final
Experimental modelling of lateral loads on large diameter
mono-pile foundations in sand
For obtaining the degree of M.Sc. in Civil Engineering at Delft University of Technology
by
Etienne A. Alderlieste
1313169
April 2011
Committee members:
Delft University of Technology, Geotechnical Engineering
Professor : Prof. ir. A.F. van Tol (chairman)
Supervisors : Dr. ir. W. Broere
: Dr. ir. J. Dijkstra
: Dr. ir. O. Heeres
Delft University of Technology, Offshore Engineering
Supervisor : Prof. ir. C.A. Willemse, MBA
Deltares
Supervisor : Ir. H.J. Luger
Gemeentewerken Rotterdam
Supervisor : Ing. H. Brassinga
General information:
The last years several offshore wind farms have been completed. Such farms typically
consist of a few dozen wind turbines. The majority of these wind turbines is founded on
mono-piles with a diameter of 4–5 m and is designed according to standards that use the
p-y method for lateral loading conditions. However, this p-y method is not validated for
such diameter piles. Full-scale field tests and model pile tests with properly scaled stress
conditions subjected to lateral loads are scarce.
This Thesis investigates the effect of a diameter increase on the lateral bearing capacity
of 2.2 m and 4.4 m diameter mono-piles in dry sand. Static displacement controlled
and one-way cyclic force controlled model pile load tests were performed using the Delft
University of Technology geotechnical centrifuge.
Results show that the smaller 2.2 m diameter pile has a lower secant and tangent soil-
pile stiffness when compared to the 4.4 m diameter pile. In all cases the tangent stiffness
is about 50% higher than the secant stiffness. Also, a force acting higher above soil surface
yields a lower initial system stiffness. Results furthermore show that an increase in pile
diameter with a constant slenderness or L/D-ratio, relative density Id and load excentricity
e, leads to a significant increase in static lateral capacity. The secant and tangent stiffness
also significantly increase with increasing diameter.
The accumulation of lateral pile displacements as function of the number of applied
one-way cyclic lateral loads has been investigated using 500 force controlled cycles. For
an increasing number of load cycles the pile head displacements increase whilst the rate
of accumulation decreases. However, additional loading cycles, up to e.g. 100000, are
recommended.
For static loading conditions experimental results have been compared to the default
API formulation for laterally loaded piles. This default formulation shows a significant
overestimation of the initial stiffness. However, after incorporating a stress dependent
secant stiffness E50 , which has been derived from triaxial compression tests on model
sand, good agreement is found between the experimental results and the modified API
formulation for pile displacements <0.1D. For displacements of about 0.05D the modified
p-y method underestimates the lateral bearing capacity of the 4.4 m diameter pile by up
to 25
Further research with loading conditions that mimic field loading conditions and a
sample with a higher initial density are recommended. It is also recommended to im-
prove the current set-up using better load control and a larger strongbox and to perform
additional model pile tests on saturated samples.
vii
Samenvatting
ix
Those who sacrifice liberty for security, deserve neither.
Benjamin Franklin
Contents
Abstract vii
Samenvatting ix
1 Introduction 1
1.1 Offshore wind energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Wind turbine foundations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Mono-pile foundation loads and soil response . . . . . . . . . . . . . . . . . 3
1.4 Design of large diameter mono-piles . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 Limitations of this research . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.7 Outline of this Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
xi
xii Table of contents
4 Test Results 31
4.1 Overview of the experimental programme . . . . . . . . . . . . . . . . . . . 31
4.2 Surface settlement and sample weight loss . . . . . . . . . . . . . . . . . . . 32
4.3 Experimental test results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3.1 Pile rotation and translation . . . . . . . . . . . . . . . . . . . . . . 33
4.3.2 Static load displacement curves . . . . . . . . . . . . . . . . . . . . . 34
4.3.3 Cyclic load displacement curves . . . . . . . . . . . . . . . . . . . . . 36
4.3.4 Secant and tangent stiffness . . . . . . . . . . . . . . . . . . . . . . . 38
4.4 Pile displacement accumulation . . . . . . . . . . . . . . . . . . . . . . . . . 39
5 Comparison 41
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 Calculation assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.3 Default API . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.4 Modified API . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.5 Stress dependent formulation . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.6 Effect of adapted stiffness on the p-y curves . . . . . . . . . . . . . . . . . . 45
5.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8 Acknowledgements 53
Bibliography 61
Appendices 65
E Model piles 89
F Pycnometer results 91
xv
xvi List of figures
C.1 Estimated CPT values for North Sea sand at the OWEZ wind farm . . . . 79
xvii
xviii
Abbreviations and Symbols
Abbreviations
xix
xx Abbreviations and Symbols
Symbols
n Porosity [-]
p Soil resistance [kN m-1 ]
pult Ultimate bearing capacity [kN]
pud Lateral soil resistance, deep [kN m-1 ]
pus Lateral soil resistance, shallow [kN m-1 ]
r Radius [m]
t Wall thickness [m]
Time [s]
V Volume [m3 ]
Vpores Volume of pores [cm3 ]
Vsolid Volume of solids [cm3 ]
Vtotal Total volume (= Vpores + Vsolid ) [cm3 ]
v Velocity [m s-1 ]
y Deflection [m]
z Depth [m]
γ Specific weight [kN m-3 ]
ε Strain [%]
ϕ Angle of internal friction [◦ ]
ω Angular velocity [rad s-1 ]
σ Stress [kN m-2 ]
σ1 Principal stress direction [kN m-2 ]
σ3 Secondary stress direction [kN m-2 ]
xxii Abbreviations and Symbols
Chapter 1
Introduction
blade(s)
tower
seabed
mono-pile foundation
In the coming decade more turbines will be installed offshore. The Dutch government
has planned a total installed generator capacity of 6000 MW for offshore sites in year 2020
[4]. Currently (2011), the total capacity is only about 200 MW. In several other countries
next to the North Sea (e.g. the United Kingdom, Belgium, Denmark and Germany) new
1
2 Introduction
offshore wind farms will also be constructed. Wind farms require large areas, O(40 km2 ),
and may not block shipping routes. Also, free space close to shore is limited. Large areas of
the North Sea are characterized by sandy soil conditions and a limited water depth. This
limited water depth allows for the construction of offshore wind farms further offshore.
Offshore generator capacity trails onshore generator capacity. In November 2010 a 7.5
MW turbine has been installed onshore in Germany by Enercon [30] whilst 5 MW turbines
are installed offshore at the Beatrice wind farm in Scotland. The Offshore Prinses Amalia
wind farm [64] consist of 60 Vestas V80-2.0 MW turbines that are founded on 4.0 m mono-
piles. The water depth is 19–24 m and the total power output is 120 MW. This wind farm
was completed in 2007 and became operational in 2008. Offshore Wind farm Egmond aan
Zee (OWEZ) [98], completed in 2006, consists of 36 Vestas V90-3.0 MW turbines located
10 to 18 km from the Dutch coast. More recently (2009), wind farms are constructed
as far as 46–52 km from shore, i.e. the Belwind [13] wind farm, comprising of 55 Vestas
V90-3.0 MW turbines situated on Bligh Bank. This farm has been completed in 2010 and
is operational. The typical water depth at Belwind and OWEZ sites ranges from 15 to 37
m.
onshore ⇐⇒ offshore
Figure 1.2 Overview of foundation types for wind turbines; from left to right: onshore piled
foundation, offshore gravity base, suction caisson, mono-pile, tripod, jacket and
floating platform foundation
level, at which the waves are loading the structure, lead to larger marine based loads.
Increasing turbine generator capacity generally requires a larger rotor, which leads to
higher towers and larger forces acting on the foundation. In order to build wind farms
with higher capacity generators further offshore, i.e. with foundations having to cope with
increasing lateral and vertical loads, the foundation dimensions have to be adapted. A
mono-pile with a larger diameter is the result.
The limited water depth and sandy soil conditions at many sites of the North Sea allow
for the construction of mono-pile foundations. An overview of some soil profiles at the
OWEZ wind farm is presented in Appendix A. The mono-pile foundation is designed for
water depths of less than 30 m. Water depths of 30 – 40 m are considered the transition
depth. For water depths >40 m jacket foundations become more economical ([1; 25; 78]).
For OWEZ and Belwind wind farms mono-piles with an approximate outside diameter of
4.4 and 5.5 m respectively have been used.
Fv
Mb
Fh
seabed
Mt
Fh
mono-pile mono-pile
Figure 1.3 Schematic overview of forces (horizontal and vertical) and moments (bending and
torsional) acting on a mono-pile foundation
Current large diameter offshore mono-piles have a slenderness ratio or embedded length
over outer diameter L/D-ratio of 5 – 7 and are considered to behave rigid. This L/D-
ratio differs significantly from the piles from which the p-y method is calibrated (L/D =
34.4) [21; 74]. For large diameter piles the initial soil stiffness response of the pile-soil
interaction will be higher compared to slender piles [8; 49; 50; 95] and the p-y method
does not explicitly take pile stiffness into account. Therefore, determination of the validity
of the p-y method for large diameter piles is necessary.
1.5 Objectives
The current research will investigate soil-pile behaviour for large diameter mono-piles sub-
jected to one-way cyclic lateral loading in medium dense sand with a consistent relative
density Id of approximately 60%. Model pile tests will be performed in a geotechnical
centrifuge and referred to as ng tests. Prototype pile diameters of 2.2 m and 4.4 m are
examined.
Primary objective:
Secondary objectives:
- Determination of validity of the p-y method for static loading of large diameter steel
piles by means of several ng model pile tests.
- Identification of loading conditions and model scaling for ng-tests and determination
of physical boundary conditions for model piles.
- The North Sea subsoil is built-up of multiple saturated soil layers that generally
consist of medium dense to dense sand with an Id of e.g. 70–95%. Layers of clay and
peat can be present as well. The model pile tests will be performed in a homogeneous
single layer of dry sand with a consistent relative density Id of 60%. This will result
in a lower lateral bearing capacity compared to denser sand.
- Offshore wind turbines are founded in saturated soil. The use of dry sand for model
pile tests results in higher effective stresses and will result in a higher lateral bearing
capacity.
Fh Fh
ϕ e e
Fr,1 zr,2
L
Fr,2
7
8 Laterally Loaded Piles
When a pile rotates a shear resistance can be mobilized at the pile toe. The presence of
this shear resistance is indicated by numerical calculations and it is suggested to be taken
into account designing large diameter mono-piles [12; 50]. According to the German design
standard GL the rotation of a laterally loaded mono-pile foundation should be minimized
in order to guarantee system operation. A maximum pile rotation of 0.5◦ at mud line is
allowed [34]. For a pile with an L/D-ratio of 5 and a diameter of 4.4 m this leads to a
deflection restriction of 15.4 cm, i.e. 0.035D. This standard, however, does not allow a
lateral displacement of the pile toe. Satisfying this latter criterion requires deep installed
mono-piles, e.g. up to 43 m for pile diameters of 5.0 m (i.e. with an L/D-ratio of 8.6)
[3], which would make this type of foundation less economical. An increase in embedded
length of a cyclic laterally loaded pile leads to a smaller rate of accumulated displacement
and therefore, piles with a higher L/D-ratio can cope better with a large number of cyclic
lateral loads [2; 3]. Mono-piles with an L/D-ratio of 5–6 have been installed offshore,
which suggests that in practice a small pile toe displacement is nevertheless accepted when
designing a laterally loaded mono-pile foundation.
Elasticity method The elasticity method takes soil continuity into account, but the soil
modulus is assumed elastic and varies with stress level [10; 70; 72]. As soil behaviour is
more accurately described in a non-linear manner, this method is limited to small strains
(i.e. smaller than 1%) and not suitable to determine ultimate lateral pile loads.
Limit state method The limit state method can determine the ultimate lateral pile
capacity pult [19]. This method, also known as Broms’ method, assumes a rigid pile and
a linear relation between pile diameter and ultimate lateral soil capacity. Finite Element
(FE) calculations indicate this limit state method underestimates the pult for sand [32].
Finite Element Method (FEM) Finite Element Analysis (FEA) is a useful tool cap-
able of modelling soil continuity, nonlinearities, complex pile-soil interaction and 3D bound-
ary conditions. Large diameter mono-piles with typical wall thickness and flexural stiffness
embedded in a sandy soil can nowadays be modelled, see e.g. [3; 12; 50]. However, FE
analysis is still mainly used for research purposes using simplified soil models. FEA can
nevertheless help improve understanding of pile-soil interaction.
Subgrade reaction method This method, which includes the p-y method, is based on
a beam on elastic foundation (BEF) and is widely used for its simplicity and reasonable
accuracy [16; 57; 75; 87]. The soil resistance is assumed to be linear. The p-y method
does take the non-linear relation between load F and pile deflection y into account using
so-called p-y curves. This subgrade reaction method is limited by the fact that soil resist-
ance is modelled as a finite number of springs, whilst soil properties should be considered
continuous. Also, the horizontal subgrade modulus is a model parameter rather than a
fundamental soil property.
The p-y method has been applied for many years and almost all large diameter offshore
mono-pile foundation dimensions are based on current design standards [6; 26; 34] which
incorporate the p-y method, despite several known limitations and uncertainties. These
limitations will be discussed in Section 2.3.
Plans for new wind farms already exist and currently mono-piles are the recommended
foundation type. Due to this increase of application of mono-piles for offshore wind farms
the research topic of soil resistance-pile deflection (p-y) is revived, especially for application
of large diameter piles with low slenderness ratios.
For sand this p-y method was derived from field tests on Mustang Island (USA) [21; 74].
These field tests consisted of 2 static and 5 cyclic load tests applied to two 0.61 m (24 inch)
diameter D piles with wall thickness t of 95 mm (3/8 inch), length L of 21 m (69 feet) and
an L/D-ratio of 34.4. The wall thickness over diameter t/D-ratio equated to about 64.
Based on the same data it has been concluded that a hyperbolic curve is an improvement
over the originally formulated expression [60]. This adapted formulation has been accepted
by several design standards e.g. [6; 26; 34]. However, in both variants multiple soil layers
and non-constant pile diameters are not accounted for. The soil-wedge (SW) model [9; 62],
which is used to predict the response of a laterally loaded flexible pile, does incorporate the
pile properties and is able to handle multiple soil layers. For sand the lateral soil resistance
versus deflection relationship at depth z is approximated by the expression,
µ ¶
kz
p(z) = A · min (pus (z), pud (z)) tanh y . (2.6)
A · min (pus (z), pud (z))
The p-y method is known to have several limitations. For instance, this method is
only suitable for constant pile diameters installed in single layered soils. Secondly, the pile
(flexural) stiffness EI is not taken into account. During driving of a mono-pile the initial
stress state and soil density are altered. For example, the ratio between the horizontal and
vertical stress component, or the lateral earth pressure coefficient K0 , is likely to change.
The API [6] assumes a constant K0 = 0.4. Other values like K0 = 1 - sinϕ have also been
used, e.g. [16]. For (medium) dense sand with a ϕ equal to 37◦ there is no difference.
4
x 10
2
s, 18 m
s, 11 m
1.5 s, 4 m
c, 18 m
c, 11 m
p [kN/m]
c, 4 m
1
0.5
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Deflection [m]
Figure 2.3 Static (s) and cyclic (c) p-y curves for sand at different depths
Numerical calculations show that application of p-y curves probably leads to an over-
estimation of the stiffness of large diameter mono-piles in sand with respect to horizontal
loading. This may lead to insufficient pile design lengths [3; 50; 95]. EI variations of the
pile do not significantly influence p-y curves for sand [32]. 3D FEA using linear elastic
soil models shows an influence of pile diameter on subgrade reaction modulus [8; 40]. This
analysis also states this influence is expected to be negligible as pile stiffness increases
with increasing diameter (the pile diameter used is 1.2 m). Non-linear behaviour of soil
tends to increase the pile diameter effect in relation to the pile response and standard p-y
curves show a tendency to overestimate soil resistance for large diameter piles for weakly
cemented sands [8; 40]. In contrast, a linear dependency of pile diameter on soil resistance
suggests no diameter dependency [51]. Applicability of the p-y method for large diameter
mono-piles therefore requires additional investigation, e.g. into the diameter influence on
the lateral bearing capacity.
mono-pile
σ3
Fh σ1 σ1
σ3 TC/TE
σ1
τ
DSS
σ1
Figure 2.4 Schematic view of laterally loaded mono-pile and soil tests: triaxial compres-
sion/extension (TC/TE) and direct (simple) shear (DSS)
and c1 > c2 was found [85]. A c2 > c1 was found investigating medium coarse and fine
sands using triaxial tests [36]. A near-logarithmic relation for up to 104 cycles and an
over-proportional logarithmic relation for a larger number of loading cycles N was found
using saturated drained triaxial tests and multi-axial direct simple shear tests [93]. Work
from latter authors also states the strain accumulation rate is not dependent on the applied
frequency (for the tested range of 0.05 – 2 Hz) and there is no significant difference between
dry and saturated drained triaxial test results. Soil type and soil density influence the strain
accumulation rate, although an increase in number of cycles N leads to an accumulation
of displacements nevertheless.
Hettler (Eq. 2.7) performed triaxial tests on dry sand and model pile tests using sand
and suggested a relation between displacement for 1 cycle w1 and the displacement for N
cycles wn depending on a material parameter Cn (for sand equal to approximately 0.2) [37].
Little and Briaud (Eq. 2.8) suggested a power relation based on the number of cycles N
[55]. Experimental investigation into the influence of regular and irregular cyclic loading
on dry granular material yields a displacement curve versus the logarithmic number of
cycles N also according to an N m relation [66].
wn = w1 + (1 + Cn ln N ) , (2.7)
wn = w1 N 0.136 . (2.8)
Undisturbed fine grained soil samples subjected to cyclic excitations show a reduction of
stiffness [65], which corresponds to the p-y method. This contradicts with results showing a
stiffness increase with increasing number of loading cycles [42; 53]. Although soil behaviour
depends on the stress state and e.g. the followed stress path for triaxial tests with cyclic
soil loading, nevertheless the p-y method and other soil-pile related topics require further
investigation.
In order to determine the validity of the p-y method for large diameter mono-piles and
the effect of diameter increase on the lateral bearing capacity, reference piles are intro-
duced. Based on the reference piles two model piles are made that will be tested using
the geotechnical centrifuge of Delft University of Technology. These ng experiments are
introduced in Section 3.9.
17
18 Experimental Test Set-up
4.4 m and 2.2 m diameter mono-piles is given in Table 3.1. In Section 3.4 the scaling laws
by which prototype diameter piles are converted to model piles are introduced. In Section
3.5 the model piles will be presented.
These Equations rely on the pile diameter D in m, soil density γ in kN/m3 , soil para-
meters Nq and β, as well as the embedded pile length L in m. The Nq value is based
on medium dense to dense sand. According to the API [7] the maximum predicted shaft
resistance and tip resistance for the reference piles respectively are 67 kPa and 3.0 MN.
The calculated total vertical bearing capacity Fbc for the 2.2 m and 4.4 m diameter
piles is 2.5 MN and 17 MN respectively. An entire OWEZ wind turbine resembles a
weight of approximately 600 ton, i.e. 6 MN, and is founded on a 4.4 m diameter mono-
pile. Therefore, the vertical bearing capacity is not the limiting factor in the design of an
offshore wind turbine mono-pile foundation.
9 70
8
60
7
50
6
pult [MN]
pult [MN]
5 40
4 30
static, e = 0.0 static, e = 0.0
3
static, e = 2.4 20 static, e = 2.4
2 static, e = 4.8 static, e = 4.8
cyclic, e = 0.0 cyclic, e = 0.0
10
1 cyclic, e = 2.4 cyclic, e = 2.4
cyclic, e = 4.8 cyclic, e = 4.8
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
Normalized lateral deflection y/D [−] Normalized lateral deflection y/D [−]
The ultimate lateral bearing capacity pult of the pile is referred to as the load cap. A
high initial stiffness and plateau for y ≥ 0.1D are visible in the load-displacement results
as shown e.g. in Figure 3.1 and Appendix D. The load-displacement results of 4.4 m and
2.2 m diameter piles embedded in dry sand are shown in Figure 3.1. The results indicate
that the mono-pile behaves rigid, as only marginal bending of the pile occurs. The centre
of rotation lies about 4D below soil surface level, which is in agreement with [16; 20]. The
normalized lateral deflection is measured at 0.0 m, 2.4 m and 4.8 m. Detailed MPile results
are included in Appendix D. As shown, an increase in load height to 2.4 m above seabed
level leads to a decrease in pult of about 13% for the 4.4 m diameter pile. A load offset of
4.8 m above seabed level leads to an additional 11% decrease of the pult .
Scaling to model space has to be done with care, for not all properties scale in a linear
manner. For these tests the main objective for the model piles is a correctly scaled flexural
stiffness. The aim is to scale the geometry as good as possible.
Void ratio and porosity The extreme void ratios are determined according to the
Japanese Geotechnical Society (JGS) [39]. To get the lowest void ratio or the highest soil
density the specified volume is filled layer by layer and gently densified. The highest void
ratio or lowest density is achieved by slowly and carefully pouring sand through a funnel
into the specified volume.
Although effort is put into producing accurate and consistent results, it is assumed that
the very extremes are not possible to be achieved in relative short amounts of time, i.e. up
to 20 minutes. The average of the 3 peak values for either the lowest density and highest
density are taken. The porosity n, void ratio e and relative density Id are determined
based on these two sets of three values. The nmin and nmax differ about 10% which is
acceptable for fine grained sand. The desired Id of 60% leads to a porosity and void ratio
of 0.394 and 0.651 respectively. The particle density is 2.6457 Mg/m3 and the specific
density is 1.603 Mg/m3 .
Vpores
n= , (3.3)
Vtotal
Vpores n
e= = , (3.4)
Vsolid 1−n
emax − e
Id = . (3.5)
emax − emin
Triaxial tests The soil particles in oven dry as well as saturated drained conditions have
been tested using a triaxial apparatus according to the British Standard Institution (BSI)
[17] and Nederlands Normalisatie-instituut (NEN) [61]. The effective confining pressure
applied is 500 kPa, which is equal to about 31 m below soil surface. Since all tests have
been performed at the same (effective) confining pressure it is not possible to determine
the cohesion. The cohesion is assumed to be 0 kPa for this granular material. In Appendix
H the deviatoric stress is plotted against the strain. Additionally, several Mohr circles are
plotted. The peak angle of internal friction ϕ equals about 35◦ . The secant stiffness of the
initial loading branch is 1400 ± 100 kN/m. The reloading stiffness is 2200 ± 100 kN/m.
Sieving The sand particles have been sieved in order to determine the particle size
distribution and coefficient of uniformity. The D10 , D50 and D60 are equal to 170, 245 and
260 µm respectively, and as the coefficient of uniformity indicates, the soil is considered
poorly graded,
D60 260
Cu = = = 1.5 ≤ 2.0. (3.6)
D10 170
100
90
80
Cumulative percentage [%]
70
60
50
40
30
20
10
0
0 50 100 150 200 250 300 350 400 450 500
Sieve diameter [µm]
Sand in open air The sand does attract almost no water from the air. Oven dry sand
that has been laying in the lab exposed for 48 hours and >480 hours resulted in a weight
increase of about 0.02% ± 0.01% and 0.03% ± 0.01% respectively. These percentages are
considered negligible and the sand is considered dry.
Several soil parameters have been determined and the experimental programme can now
further be accommodated. The Delft University of Technology geotechnical centrifuge, in
which the ng experiments will be performed, is introduced next. The sample preparation
and experimental programme will be introduced respectively in Sections 3.8 and 3.9.
mv 2
F = = mω 2 r, (3.7)
r
v = ωr. (3.8)
y v F
swing
y(θ, t)
r
dθ
dt
=ω
y′ x′
θ
x
0 x(θ, t)
Figure 3.3 Global polar coordinates of the model and local coordinate system of the model
A control room housing 3 PCs is situated next to the room in which the centrifuge
set-up is located. Using one of the PCs the tangential speed (i.e. the acceleration) can
be set and controlled. The connection between the PC, motor and gearbox involves a
controller. The revolutions of the beam and temperature of the bearings are monitored.
During the tests performed for this Thesis the revolution monitoring was not operational.
A second PC is used to send data via slip rings to the mini-ITX PC system mounted
on the beam of the centrifuge. This is also possible during full centrifuge operation. This
connection consists of two times 100Mbit and is also used to transfer other data, e.g. from
the 5 mega pixel (MP) camera mounted on the swing. An overview of the entire set-up is
shown in Figure 3.6.
horizontal and vertical movement are controlled. A National DAQ card is used to convert
the analogue signal from load cells (for both vertical as well as horizontal loading) to a
digital signal. Data is then stored on the 64 GB solid state disk (SSD) attached to the
PC. If desired, this data is directly accessible using a PC connection via the slip rings. A
more detailed overview of equipment is shown in Figure 3.6.
The horizontal movement of the load frame is measured at a rate of 5 Hz. Since
the speed is 0.5 mm/sec this leads to an accuracy of approximately 0.1 mm. The load
corresponding to the lateral movement of the load frame is registered at 16 Hz. The
maximum available load on the load cell is 1500 N, which at 160g equates to 38 MN.
CENTRIFUGE
Fh TA
soil surf ace e
B1
L
170 180
sand model pile
B2 D
238
Strongbox dimensions The strongbox is a metal box with internal length, width and
height of 238, 222 and 180 mm respectively, and of which an overview is given in Figure
3.7. The weight of the empty box is 5401 g ± 1 g. In loading direction the total space
is 238 mm. The pile with outer diameter D will be installed in the strongbox up to the
required depth L = 5D. The space available between the horizontal boundary B1 and
vertical boundary B2 is in the order of 3.7D and 1D respectively for the large model pile.
For the small model pile B1 and B2 equal 9D and 6D respectively. The force F applied
to the pile and measured at height e above soil level. The displacement of the pile is also
measured at location T A, equal to e above the soil level.
Sample preparation For each test a new sample was prepared by pouring dry sand
in the strongbox and subsequently vibrating the entire sample. This method resulted
in consistent soil densities with relative densities Id varying between 58–62% for all tests.
After initial sample preparation the model pile was installed at 1g and the entire strongbox
was again briefly vibrated to minimize pile installation effects. The total weight of the
sample is determined and the sample is carefully transported to the centrifuge room. The
total weight of about 20 kg does not require special tools for transportation. The initial
pile positions before flight were measured and the sample is placed on one of the swings of
the centrifuge.
displacement
pult
time
0
(a) Static loading scheme
load
pult
load cap
cycles
0 1 2 3 4
(b) Cyclic loading scheme
F1 − F2 F1
Esec = = , (3.9)
u1 − u2 u1 − u2
F4 − F3
Etan = . (3.10)
u4 − u3
Load F
Etan Esec
F1
F4
F3
F2 Displacement u
0 u2 u3 u4 u1
Model pile installation The ratio between model pile wall thickness t and grain size
D50 is limited, about two for the small diameter model pile and 4 for the large diameter
model pile. Installing the model piles at 1g limits the build-up of high stresses below the
pile tip compared to pile installation at ng. Also, since the model piles will not be loaded
by forces other then the self-weight and the main force is acting laterally, the influence on
the lateral pile response of this low t/D50 ratio is considered negligible.
Shear band width and particle size versus model wall thickness and diameter
The shear band width is relevant for centrifuge experiments as often a low ratio between pile
diameter and grain size is found. Shear band widths are about 16D50 [59; 83]. Furthermore
a value of > 100D50 is desired for the pile diameter. In this research a shear band width
and suggested pile diameter of 3.9 mm and 24.5 mm respectively are found. Since the
governing pile (failure) mechanism is rotation, which leads to shear below the pile toe, the
clearance between pile toe and strongbox bottom of >1D should be sufficient.
Void ratio decrease Due to the applied acceleration field the sample possible could
densify. Tests with an initial porosity of 39.7% to 48.4% and acceleration of 120g indicate
a decrease in overall sample porosity of approximately 1% [83]. The majority of this
densification will have occurred before the tests started. The current set-up does not allow
for measuring a porosity decrease during flight. Nevertheless, soil levels will be measured
before and after flight and the results will be presented in Chapter 4.
Soil density The Id of 60% is low compared to North Sea soil conditions for several
locations, but makes the density of the sample better controllable and consistent and it
should deliver reproducible results. This relative low density will lead to a lower (initial)
stiffness of the soil.
Loading conditions The current set-up is designed for one-way loading of a model pile
foundation at a single height. Different heights at which forces in the field act (i.e. wind
loads act at a different height than marine loads) cannot be modelled using the current
set-up. Neither is it possible to load the foundation from multiple directions.
All ng experiments are conducted on dry sand, using static or one-way cyclic loading
conditions with a force at a fixed height for each experiment. In reality loading conditions
are not that simple, as forces from different directions, with varying magnitude and at
different heights load the wind turbine. Also, since pile tests are performed using dry
sand, the effective stresses are higher compared to saturated sand.
Test Results
The results from model pile tests, of which an overview is given in Table 4.1, will be
presented in Section 4.3. The load excentricity e, number of load cycles N and level of
acceleration n are shown. The EI represents the flexural stiffness of the corresponding
model pile, of which the prototype diameter is indicated by Dp .
a 2.2 5 0 1 160 35
b 2.2 5 4.8 1 160 35
c 2.2 5 0 1 80 558
d 2.2 5 2.4 1 80 558
e 2.2 5 4.8 1 80 558
f 2.2 5 2.4 500 80 558
g 4.4 5 0 1 160 558
h 4.4 5 2.4 1 160 558
i 4.4 5 4.8 1 160 558
j 4.4 5 2.4 500 160 558
k 4.4 5 4.8 500 160 558
In order to better appreciate the results presented in this Chapter, first soil surface set-
tlements and the mass difference of the sample before and after experiments are discussed.
Thereafter, in Section 4.3 pile load-displacement results and the effect of load excentricity
will be presented.
31
32 Results
No additional measurements have been performed to quantify the rate of particle settle-
ment. However, typically the major part of the settlements occurs within the first minutes
of flight [83]. The limited number of measured points only gives an indication whilst e.g.
laser scanning of the surface could create a more accurate view of the settlements. When
measuring the surface settlement during flight also the settlement rate can be quantified.
The mean for test 1 and 2 is equal to surface settlement respectively of 0.26 ± 0.3
mm and 0.22 ± 0.3 mm. This relatively high uncertainty, which is in the same order of
magnitude as the measured settlement, is the result of the manual measurement method.
Nevertheless, a general trend is distinguished and no surface rise has been witnessed (ob-
viously, except on the passive side of the laterally loaded pile).
A surface drop of 0.3 mm results in a volume decrease of 15.7 cm3 or 0.2% and an Id
increase from the desired 60% to 60.7%. The influence of surface settlements on the pile
load tests therefore is limited.
Sample weight loss Before and after each test the weight of the strongbox, including
sand and model pile, has been measured. The strongbox and model pile will not loose
mass during the experiments. Therefore, the weight difference is due to the loss of sand
particles. The average amount of sand particles lost during flight is less than 1 gram, which
makes the total mass loss negligible. The running time of tests has no effect on the total
amount of mass loss.
model pile
c
Fh
b
a
Figure 4.1 Schematic side view of initial pile position (continues line), displaced pile position
(dashed line) and measurement locations a, b and c
To verify that no deviation of the pile from the loading direction occurs, at 5 mm and
12 mm above soil surface level the pile position before and after experiments is examined
perpendicular to the loading direction. Sideways pile displacements are <0.6 mm ± 0.2
mm for all recorded cases. The pile thus displaces in line with the load frame, i.e. in the
prescribed loading direction.
The initial and final model pile positions have been measured at 1g. Therefore, the
exact position of the model pile in the final state during flight remains unknown. More
over, even when the load frame is stopped during lateral loading of the model pile and the
system is spun down, the soil is unloaded, thereby influencing the pile position.
Constructing the pile rotation angle from differences between the measured locations
a–b, b–c and a–c, the centre of rotation is found to be between 4D and about 6D. Any
value >5D suggests pile translation. Although the lateral pile displacement is measured
sufficiently accurate, the distance between the positions a, b and c is too small to properly
distinguish between rotation and translation. When taking the centre of rotation at 4D
below soil surface, the average pile angle is consistent. However, measurement errors are
smeared out over a length of approximately 4.5D. In order to distinguish between pile
translation and rotation and to better quantify the pile rotation, measurement of pile
displacements have to be significantly improved.
0.12 0.12
0.1 0.1
Load/(L⋅D2) [MN/m3]
Load/(L⋅D2) [MN/m3]
0.08 0.08
It was not possible to perform two model pile test using the same sample. Therefore,
minor inconsistencies in soil density between the two samples may be present. The small
13.68 mm and large 27.36 mm diameter model piles were located at respectively 8.5Ds and
3.7Dl from the strongbox boundary. However, ideally a much larger strongbox is required
to be certain a negligible influence of strongbox boundary on the model pile test results
is present. Also, a minimal pile diameter of 24.5 mm is recommended for the current D50
of 245 µm [59; 79; 83]. Nevertheless, good agreement between both piles is visible for pile
displacements up to 0.25D. For displacements up to 0.25D the influence of the strongbox
boundary is limited and the use of the large 27.36 mm diameter model pile is acceptable.
0.08
0.07
0.06
Load/(L⋅D2) [MN/m3]
0.05
0.04
0.03
0.02
- A stiffness increase during the first 10 cycles is visible. Later cycles show only minor
fluctuations in secant and tangent stiffness, see Figures 4.8a and 4.8b.
- A 0.2 s time gap between the displacement measurement and force measurement is
present. When the target load is reached the load frame continues movement for 0.2
s, thereby further displacing the pile, which results in a higher lateral load. However,
as the secant and tangent stiffness do not significantly change for >50 cycles, this
should lead to a constant loading level.
- During tests no airconditioning was present. As result, during longer tests of e.g.
two hours the room temperature increased from about 20◦ to >30◦ , which may have
had an effect on the response of the load cell. Eventhough the force plotted against
the applied number of cycles shows only a marginal increase of the force for N>50
cycles, see Figure 4.6.
2 14
12
1.5
10
Load [MN]
Load [MN]
8
1
6
4
0.5
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
y/D [−] y/D [−]
Figure 4.4 Load-displacement curves for one-way cyclic laterally loaded piles
0.2 0.2
0.15 0.15
y/D [−]
y/D [−]
0.1 0.1
0.05 0.05
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Number of cycles [−] Number of cycles [−]
0.2 0.2
0.15 0.15
y/D [−]
y/D [−]
0.1 0.1
0.05 0.05
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Number of cycles [−] Number of cycles [−]
(c) 2.2 m diameter, e = 2.4 m, the first 25 cycles (d) 4.4 m diameter, e = 2.4 m, the first 25 cycles
0.2 0.2
0.15 0.15
y/D [−]
y/D [−]
0.1 0.1
0.05 0.05
0 0
475 480 485 490 495 500 475 480 485 490 495 500
Number of cycles [−] Number of cycles [−]
(e) 2.2 m diameter, e = 2.4 m, the last 25 cycles (f ) 4.4 m diameter, e = 2.4 m, the last 25 cycles
2 15
1.5
10
Force [MN]
Force [MN]
1
5
0.5
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Number of cycles [−] Number of cycles [−]
(a) 2.2 m diameter, e = 2.4 m, load F (N ) (b) 4.4 m diameter, e = 2.4 m, load F (N )
Load F
Etan Esec
F1
F4
F3
F2 Displacement u
0 u2 u3 u4 u1
A smaller pile diameter shows a lower secant and tangent stiffness. In all cases the
tangent stiffness is about 50 % higher than the secant stiffness. Also, larger excentricities
e yield a lower stiffness. The results show that an increase in pile diameter with constant
L/D, Id and e, results in a significant increase in static lateral capacity and secant and
tangent stiffness from cyclic load tests.
90 90
D = 4.4 m, e = 2.4 m
80 D = 4.4 m, e = 4.8 m 80
D = 2.2 m, e = 2.4 m
70 70
60 60
50 50
D = 4.4 m, e = 2.4 m
40 40
D = 4.4 m, e = 4.8 m
30 30 D = 2.2 m, e = 2.4 m
20 20
10 10
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Number of cycles [−] Number of cycles [−]
0.25
D = 2.2 m, e = 2.4 m
Normalized pile displacement y/D [−]
D = 4.4 m, e = 2.4 m
0.2 D = 4.4 m, e = 4.8 m
Hettler (w = 0.05)
1
Little & Briaud (w1 = 0.05)
0.15 Hettler (w = 0.10)
1
Little & Briaud (w = 0.10)
1
0.1
0.05
0 0 1 2
10 10 10
Number of cycles [−]
The empirical formulae do not correspond well to the soil behaviour from the model
pile test data. Model pile test results show an over-logarithmic rate of stiffness increase
for the first 100 cycles and an under-logarithmic rate thereafter.
Two tests with an excentricity of 2.4 m show good agreement and the e = 4.8 test
shows a similar accumulation trend, but with a higher initial offset. A higher number of
load cycles is required to better understand soil behaviour. It is further recommended to
investigate the effect of varying load magnitude on the pile displacement accumulation.
Comparison
5.1 Introduction
In order to investigate the applicability of the API for pile diameters of 2.2 m and 4.4 m,
the results of the model pile tests presented in Chapter 4 will be compared with the design
calculations. For this, the soil properties and initial test conditions presented in Chapter
3 will be used as input for the API method as presented in Chapter 2 and Appendix B.
First the default formulation will be adopted before going to a modified version with a
stress dependent expression for the soil stiffness.
41
42 Comparison
0.3 0.3
k = 44000, e = 0.0 m k = 44000, e = 0.0 m
k = 44000, e = 2.4 m k = 44000, e = 2.4 m
0.25 k = 44000, e = 4.8 m 0.25 k = 44000, e = 4.8 m
experiment, e = 0.0 m experiment, e = 0.0 m
Load/(L⋅D2) [MN/m3]
Load/(L⋅D2) [MN/m3]
0.2 experiment, e = 0.0 m 0.2 experiment, e = 2.4 m
experiment, e = 2.4 m experiment, e = 2.4 m
experiment, e = 2.4 m experiment, e = 4.8 m
0.15 experiment, e = 2.4 m 0.15 experiment, e = 4.8 m
experiment, e = 4.8 m experiment, e = 4.8 m
0.1 0.1
0.05 0.05
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Normalized pile displacement y/D [−] Normalized pile displacement y/D [−]
(a) D = 2.2 m (b) D = 4.4 m
Figure 5.1 Comparison of experimental results and lateral bearing capacity conform the API
for default soil stiffness k = 44000 kN/m3
0.3 0.3
k = 1400, e = 0.0 m k = 1400, e = 0.0 m
k = 1400, e = 2.4 m k = 1400, e = 2.4 m
0.25 k = 1400, e = 4.8 m 0.25 k = 1400, e = 4.8 m
experiment, e = 0.0 m experiment, e = 0.0 m
Load/(L⋅D2) [MN/m3]
Load/(L⋅D2) [MN/m3]
0.2 experiment, e = 0.0 m 0.2 experiment, e = 2.4 m
experiment, e = 2.4 m experiment, e = 2.4 m
experiment, e = 2.4 m experiment, e = 4.8 m
0.15 experiment, e = 2.4 m 0.15 experiment, e = 4.8 m
experiment, e = 4.8 m experiment, e = 4.8 m
0.1 0.1
0.05 0.05
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Normalized pile displacement y/D [−] Normalized pile displacement y/D [−]
(a) D = 2.2 m (b) D = 4.4 m
Figure 5.2 Comparison of experimental results and lateral bearing capacity conform the API
with adapted soil stiffness k = 1400 kN/m3
−5
−10
Depth [m]
−15
−20
−25
−30
−35
0 500 1000 1500
k value [kN/m3]
³ zγ ´n
k(z) = k500kPa (5.1)
500kPa
Where k(z) in kN/m3 is the depth dependent soil stiffness, k(500kP a) in kN/m3 is the
stiffness measured at the reference stress level of 500 kPa, z in m the depth and γ the
specific weight in kN/m3 . For sand the exponent n = 0.5 can be taken. This modification
improved the quality of the API p-y method considerably, as shown in Figure 5.4.
0.25 0.15
k(z), e = 0.0 m k(z), e = 0.0 m
k(z), e = 2.4 m k(z), e = 2.4 m
0.2 k(z), e = 4.8 m k(z), e = 4.8 m
experiment, e = 0.0 m experiment, e = 0.0 m
Load/(L⋅D2) [MN/m3]
experiment, e = 0.0 m 0.1 experiment, e = 2.4 m
experiment, e = 2.4 m experiment, e = 2.4 m
Load [MN]
0.15
experiment, e = 2.4 m experiment, e = 4.8 m
experiment, e = 2.4 m experiment, e = 4.8 m
0.1 experiment, e = 4.8 m experiment, e = 4.8 m
0.05
0.05
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
Normalized pile displacement y/D [−] Normalized pile displacement y/D [−]
(a) D = 2.2 m (b) D = 4.4 m
0.05 0.05
k(z), e = 0.0 m k(z), e = 0.0 m
k(z), e = 2.4 m k(z), e = 2.4 m
0.04 k(z), e = 4.8 m 0.04 k(z), e = 4.8 m
experiment, e = 0.0 m experiment, e = 0.0 m
Load/(L⋅D2) [MN/m3]
Load/(L⋅D2) [MN/m3]
0.01 0.01
0 0
0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1
Normalized pile displacement y/D [−] Normalized pile displacement y/D [−]
(c) D = 2.2 m - detail (d) D = 4.4 m - detail
Figure 5.4 Comparison of experimental results and lateral bearing capacity conform the API
with adapted soil stiffness k as function of depth z
10000 10000
z = 11 m, k = 44000 z = 11 m, k = 44000
z = 11 m, k = 1400 z = 11 m, k = 1400
8000 z = 11 m, k(z) 8000 z = 11 m, k(z)
z = 5 m, k = 44000 z = 5 m, k = 44000
z = 5 m, k = 1400 z = 5 m, k = 1400
z = 5 m, k(z) z = 5 m, k(z)
6000 6000
p [kN/m]
p [kN/m]
4000 4000
2000 2000
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Normalized pile displacement y/D [−] Normalized pile displacement y/D [−]
5.7 Conclusions
The results of the experiments have been compared with the calculated bearing capacity
from the default and modified API method. The modified API method has a correction
term for the stress dependency of the soil stiffness. The initial stiffness response is largely
over predicted with the default relation between initial soil density and soil strength. The
results already dramatically improve if the secant stiffness E50 from the laboratory tests
is used instead. Even better predictions are obtained if the latter stiffness is adapted to a
stress dependent, i.e. depth dependent, stiffness relation of [29]. For the 2.2 m diameter pile
the p-y method in current adapted form underestimates the lateral bearing capacity by up
to 50% for displacements up to 0.1D. Displacements of the 4.4 m diameter pile up to 0.1D
can be predicted with about 25% accuracy. Differences in results between the p-y method
for both 2.2 m and 4.4 m diameter piles underlines the fact that further research into the
applicability of the p-y method and diameter dependency of lateral bearing capacity is
needed.
For the force controlled cyclic ng experiments, displacements of about 0.25D were pre-
scribed. Mud line displacements from available OWEZ field data will be compared to soil
surface displacements of model pile tests.
6.1.1 Assumptions
The bolted connection between tower and transition piece is able to transfer bending and
torsional moments and is located 11.7 m above sea level. The grouted connection between
transition piece and mono-pile foundation is assumed to be able to transfer moments as
well. The accelerometer is installed about 34 m above the mud line. Since the centre of
rotation is present at 4D below mud line level, the mud line displacement is about 0.5
times the accelerometer displacement.
The period analysed is 20–24 November 2008. On the 20th and 21st the wave height and
wind speed respectively were in the order of 5–7 m and 15–20 m/s. The other days more
calm wind and sea conditions were present.
47
48 OWEZ Field Data Analysis
−3 −4
x 10 x 10
0.8 2
|Y(f)|
|Y(f)|
0.6
0.4 1
0.2
0 0
0 4 8 12 16 20 24 28 32 0 0.25 0.50 0.75 1.0
Frequency [Hz] Frequency [Hz]
(a) Accelerations, NS (b) Displacements and velocities, NS
Displacements The total maximum displacement at mud line level during the 20th and
21st of November was 9 cm from the origin, which is about 0.02D. The information shown
in Figures 6.3a and 6.3b corresponds to a height of 34 m above mud line level.
−3 −4
x 10 x 10
0.8 2
|Y(f)|
|Y(f)|
0.6
0.4 1
0.2
0 0
0 4 8 12 16 20 24 28 32 0 0.25 0.50 0.75 1.0
Frequency [Hz] Frequency [Hz]
(a) Accelerations, EW (b) Displacements and velocities, EW
The ng model pile test displacements of up to 0.2D are about 10 times larger than the
0.02D found for field loading conditions as present on the 20th and 21st of November 2008.
Higher wind speeds or higher waves can occur and therefore the lateral pile deflection at
mud line can be larger, e.g. up to 0.03–0.04D, which still is in agreement with the maximum
0.1 0.1
Velocity [m/s] Velocity [m/s]
Displacement [m] Displacement [m]
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time [s] Time [s]
(a) NS (b) EW
pile rotation of 0.5◦ (for a 4.4 m diameter pile with an L/D-ratio of 6) as suggested by the
GL [34].
The field displacements occur at a frequency of 0.01–0.4 Hz and about 0.1 and approx-
imately 0.4 Hz are peak frequencies. Loads with a period of 3–8 s, i.e. 0.333–0.125 Hz, are
caused by waves. The operational and maximum rpm are 16.1 and 18.4 respectively, which
leads to 0.268 Hz and 0.301 Hz respectively. The effect of a blade temporarily shielding
the tower from wind would lead to frequencies of 0.81 and 0.92 Hz respectively and is
negligible compared to the total movement of the tower or foundation for the analysed
period and loading conditions.
Although the currently available data does not distinguish between wind and marine
loads, it is possible to determine pile displacements under different conditions. A calm sea
state and <2 Bft winds will result in small pile movements and displacements. A rough
sea state with high waves and high wind speeds will result in larger pile displacements.
After analysing pile displacements for several loading conditions a better (extrapolated)
prediction can be made for pile displacements under severe sea state conditions. This is an
important field of research as small amplitude cyclic loads with a relative high frequency
of about 0.3 Hz might lead to pore pressure build-up and low frequency waves of about
0.1 Hz are expected to have a much greater effect on pile displacements. It is therefore
recommended to accommodate an experimental programme with the pile displacements as
derived from these field data.
Conclusions and
Recommendations
7.1 Conclusions
In current engineering practice >2 m diameter mono-piles are designed using design stand-
ards that rely on the p-y method for lateral loading. This p-y method was derived from
0.61 m diameter 21.0 m long slender piles (L/D-ratio = 34.4) and has not been validated
for rigid piles with a diameter >2 m. In order to investigate the validity of the p-y method
for large diameter rigid mono-piles a new test setup for the geotechnical centrifuge has
been designed. The setup is capable of laterally loading of model piles using displacement
or load control. Both static and cyclic loads up to 0.1 Hz can be applied. In addition two
prototype diameter mono-piles with an L/D-ratio of 5 and diameters of 2.2 m and 4.4 m
have been scaled to model dimensions with according flexural stiffness whilst keeping a
reasonable scaling of the pile geometry.
The new setup has been used to conduct 11 experiments which study the influence of
pile diameter and load excentricity on the lateral bearing capacity and tangent and secant
stiffness response by means of static and cyclic loading schemes.
The properties of the sand have been determined in laboratory tests. The minimum
and maximum porosity are nmin = 0.324 and nmax = 0.475. Triaxial compression tests
on dry sand and saturated sand (with a Skempton B-value of 0.95) showed a peak friction
angle of 35◦ and a E50 = 1400 kPa. The reloading stiffness is 2200 kPa.
An increase in pile head displacement or pile rotation leads to an increase in lateral
static capacity. However, no ultimate lateral bearing capacity pult was found in these
experiments where pile head displacements up to 0.9D have been applied.
An increase in load excentricity leads to a reduction of static capacity and lower initial
stiffness. A load excentricity of 4.8 m above soil level compared to 0.0 m, results in
a decrease of about 20%. This reduction in capacity can be approximated by (Lemb ·
F )/(Lemb + e).
In the cyclic tests 90% of the increase in stiffness is gained in the first 20 cycles. The
other 10% gain is accumulated in the remaining 480 cycles. Furthermore, the secant and
tangent stiffness of the 4.4 m diameter pile are approximately 4 times higher than the
51
52 Conclusions and Recommendations
secant and tangent stiffness for the 2.2 m pile. The tangent stiffness for all load cases is
about 1.5 times the secant stiffness and both reach a plateau starting in the 20th cycle up
to the last prescribed cycle (500).
Comparing calculated results based on the default API formulation with experimental
results, a significant overestimation of the initial stiffness for both the 2.2 m and 4.4 m
diameter piles is found. Substituting the default API stiffness parameter k with an E50
secant stiffness based on triaxial compression tests of the model sand, a less poor fit is
obtained, where the height of the triaxial test sample is taken into account in order to
arrive at matching dimensions with the k parameter, i.e. kN/m3 . Only after incorporating
a stress dependent formulation for the stiffness parameter k does the API formulation
show good agreement with the experimental results. However, further investigation into
mono-pile-soil interaction is required.
OWEZ field data from a period in November 2008 have been analysed. The marine
conditions in the analysed period are a wave height of 5.0–6.0 m, a wave period in the
order of 6.0–8.0 s and a wind speed of about 15–18 m/s. These conditions lead to pile
displacements at seabed level up to 10 cm, i.e. about 0.02D. Displacements of the ng
experiments are much larger, but also have a lower stiffness compared to the OWEZ field
soil stiffness. A higher initial density therefore is recommended for further research.
7.2 Recommendations
To gain more insight into the soil-pile behaviour of large diameter mono-piles used as
foundation for offshore wind turbines it is recommended to perform more advanced exper-
iments.
The nacelle, which has a weight in the order of 300 ton, is located about 100 m
above the sea floor. Due to wind and wave loading the entire structure gently sways.
Experiments in which the axial load and structure geometry are better represented
will deliver a better understanding of offshore wind turbine foundations.
A greater space between strongbox boundary and pile is advised as boundary effects
for a laterally loaded pile cannot be considered absent using the current set-up.
Either a larger strongbox or additional investigation into the boundary effects are
recommended.
Waves, wind and currents load the entire structure from different directions and with
varying magnitude. Experiments with a more realistic load spectrum will improve
soil-pile interaction predictions for complex loading scenarios.
Improved lateral load control and for rapid load conditions a synchronisation between
displacement and load is required.
In order to better quantify the soil behaviour under long term cyclic loading the
number of load cycles should be increased dramatically, for instance up to 100000.
Additionally, the load spectrum should be adapted to mimic field loading conditions.
Acknowledgements
This Thesis is the final piece of the puzzle required to complete the study Civil Engineering
& Geosciences at the Delft University of Technology. The work has been carried out at
Gemeentewerken (Public Works) Rotterdam and Delft University of Technology.
During this work I have received advice from many people and I have learnt many
things about many topics, from politics to aerospace engineering.
I would like to thank my committee members for answering questions and giving me
feedback: Prof. Frits van Tol, Wout Broere, Jelke Dijkstra, Otto Heeres, Dirk Luger and
Prof. Kees Willemse. Both Otto Heeres and Henk Brassinga made it possible to perform
part of my study at Gemeentewerken Rotterdam. I had a good time with nice, interested
and interesting colleagues, thanks.
The model pile tests were carried out using the TU Delft geotechnical centrifuge. I am
grateful to Jelke Dijkstra, my daily supervisor, for giving me the opportunity to use this
rather unique piece of equipment and for his critical thoughts and healthy dose of sarcasm.
Another thanks goes out to Nuon [63] and ECN [31], from whom I received a truly
immense amount of data from two wind turbines of the OWEZ wind farm. Analysing this
the data turned out to be challenging and the results are a valuable addition to this work.
Furthermore I would like to thank many friends and acquaintances for opening their
minds to discuss many different subjects with me. And last, but certainly not least, my
parents and brother for supporting me, regardless of the situation I am in.
April, 2011
Etienne Alderlieste
53
Bibliography
[7] , Recommended Practice for Planning, Designing and Constructing Fixed Off-
shore Platforms - Working Stress Design, API RP 2A-WSD - Errata (2008), 2008.
[10] P. K. Banerjee and T. G. Davis, The Behaviour of Axially and Laterally Loaded
Single Piles Embedded in Non-Homogeneous Soils, Géotechnique, 28 (1978), pp. 309–
326.
[11] Y. O. Barton and W. D. L. Finn, Lateral Pile Response and p-y Curves From
Centrifuge Tests, in 15th Annual Offshore Technology Conference, OTC, Houston,
Texas, USA, May 1983.
55
[12] L. Bekken, Lateral Behavior of Large Diameter Offshore Monopile Foundations For
Wind Turbines, Master’s thesis, Delft University of Technology, Faculty of Civil En-
gineering and Geosciences, 2009.
[14] J. L. Bijnagte and H. J. Luger, Manual MPile - Analysis of Piles and Pile Groups,
version 4.2.3.1, 2009. Delft GeoSystems, http://www.delftgeosystems.nl.
[15] L. Brant and H. I. Ling, Centrifuge Modeling of Piles Subjected to Lateral Loads,
in Soil Stress-Strain Behaviour: Measurement, Modeling and Analysis. Geotechical
Symposium in Roma, Italy, March 2006.
[16] J. Brinch Hansen, The Ultimate Resistance of Rigid Piles Against Transversal
Forces, (1961). Akademiet for de Tekniske Videnskaber, Geoteknisk Institut, Geo-
technical Institute Denmark, Bulletin no. 12, Copenhagen.
[17] British Standard Institution, Methods of Test for Soils for Civil Engineer-
ing Purposes. Shear Strength Tests (Total Stress), BS1377, part 7, 1990, 1990.
http://www.bsigroup.com.
[19] B. B. Broms, Lateral Resistance of Piles in Cohesionless Soil, Journal of the Soil
Mechanics and Foundation Divisions, 90 (1964), pp. 123–156.
[20] N. H. Christensen, Model Tests with Transversally Loaded Rigid Piles in Sand,
(1961). Akademiet for de Tekniske Videnskaber, Geoteknisk Institut, Danish Geo-
technical Institute, Bulletin no. 12, Copenhagen.
[22] W. H. Craig, The Seven Ages of Centrifuge Modelling, (2001). Workshop at Monte
Verita on Constitutive and Centrifuge Modelling: Two Extremes.
[23] N. N. Davidenkov, The new Method of the Application of Models to the Study of
Equilibrium of Soils, Journal of Technical Physics, 3 (1933), pp. 31–136. Moscow, in
Russian.
[26] Det Norske Veritas, Offshore Standard DNV-OS-J101, Design of Offshore Wind
Turbine Structures, October 2007. http://www.dnv.com.
[27] J. Dührkop and J. Grabe, Monopilegründungen von Offshore-Windenergieanlagen
- Zum Einfluss einer Veränderlichen Zyklischen Lastangriffsrichtung, Bautechnik, 85
(2008), pp. 317–321.
[29] J. M. Duncan and C.-Y. Chang, Nonlinear Analysis of Stress and Strain in Soils,
Journal of the Soil Mechanics and Foundations Division, 96 (1970), pp. 1629–1653.
[32] C.-C. Fan and J. H. Long, Assessment of Existing Methods for Predicting Soil
Response of Laterally Loaded Piles in Sand, Computers and Geotechnics, 32 (2005),
pp. 274–289.
[34] Germanischer Lloyd, Guideline for the Certification of Offshore Wind Turbines,
2005. http://www.gl-group.com.
[39] Japanese Geotechnical Society, Soil Testing Standards, Test Methods for the
Minimum and Maximum Densities of Sands, 1996. In Japanse.
[45] C. LeBlanc, Design of Offshore Wind Turbine Support Structures - Selected Topics
in the Field of Geotechnical Engineering, PhD thesis, Department of Civil Engineering,
Aalborg University, Denmark, 2009.
[47] K. L. Lee and J. A. Focht, Cyclic testing of Soil for Ocean Wave Loading Problems,
Marine Geotechnology & Geotechnology, 1 (1976), pp. 305–325.
[49] K. Lesny, Foundations for Offshore Wind Turbines - Tools for Planning and Design,
2010. ISBN: 978-3-86797-042-6.
[50] K. Lesny, S. G. Paikowsky, and A. Gurbuz, Scale Effects in Lateral Load Re-
sponse of Large Diameter Monopiles, in Proceedings of Sessions of Geo-Denver, Den-
ver, Colorado, USA, Februari 2007. Geotechnical Special Publication no. 158.
[51] K. Lesny and J. Wiemann, Design Aspects of Monopiles in German Offshore Wind
Farms, in ISFOG, Proceedings of the International Symposium on Frontiers in Off-
shore Geotechnics, Perth, Australia, September 2005, pp. 383–389.
[53] Z. Li, S. Haigh, and M. Bolton, Centrifuge Modelling of Mono-pile Under Cyclic
Lateral Loads, in Proceedings of the 7th International Conference on Physical Model-
ling in Geotechnics, Zürich, 2010, pp. 965–970.
[54] S.-S. Lin and L.-C. Liao, Permanent Strains of Piles in Sand due to Cyclic Lat-
eral Loads, Journal of Geotechnical and Geoenvironmental Engineering, 125 (1999),
pp. 798–802.
[55] R. Little and J.-L. Briaud, Full Scale Cyclic Lateral Load Tests on Six Single
Piles in Sand, (1988).
[56] J. Long and G. Vanneste, Effects of Cyclic Lateral Loads on Piles in Sand, Journal
of Geotechnical Engineering, 120 (1994), pp. 225–244.
[57] H. Matlock and L. C. Reese, Generalized Solutions for Laterally Loaded Piles,
Journal of the Soil Mechanics and Foundation Divisions, 86 (1960), pp. 63–91.
[59] H.-B. Mühlhaus and I. Vardoulakis, The Thickness of Shear Bands in Granular
Materials, Géotechnique, 37 (1987), pp. 271–283.
[62] G. Norris, Theoretically Based BEF for Laterally Loaded Pile Analysis, in Proceed-
ings of the 3rd International Conference on Numerical Methods in Offshore Piling,
1986, pp. 361–386.
[65] D. V. Okur and A. Ansal, Stiffness Degradation of Natural Fine Grained Soils
During Cyclic Loading, Soil Dynamics and Earthquake Engineering, 27 (2007),
pp. 843–854.
[69] G. Y. Pokrovsky and I. S. Fedorov, Studies of Soil Pressures and Soil Deform-
ations by Means of a Centrifuge, in ISSMFE, 1st International Symposium on Soil
Mechanics and Foundation Engineering, Harvard, USA, vol. 1, 1936, p. 70.
[73] , Laterally Loaded Piles - Design Methods and Recent Developments, (2010).
Presentation given as part State-of-the-Art Design of Pile Foundations at Deltares,
Delft, the Netherlands.
[76] L. C. Reese and W. F. van Impe, Single Piles and Pile Grounds Under Lateral
Loading, Balkema, 2001. ISBN: 90 5809 340 90.
[78] P. Schaumann and C. Böker, Can Jackets and Tripods Compete with Monopiles,
in Copenhagen Offshore Wind, 26–28 October 2005.
[82] H. E. Stewart, Permanent Strains from Cyclic Variable Loading Amplitude Load-
ings, Journal of Geotechnical Engineering, 112 (1986), pp. 646–660.
[83] H. G. Stuit, Sand in the Geotechnical Centrifuge, PhD thesis, Delft University of
Technology, Faculty of Civil Engineering, 1995.
[85] A. S. J. Suiker, E. T. Selig, and R. Frenkel, Static and Cyclic Triaxial Testing
of Ballast and Subballast, Journal of Geotechnical and Geoenvironmental Engineering,
131 (2005), pp. 771–782.
[86] R. Taylor (ed.), Geotechnical Centrifuge Technology, Blackie, 1995. ISBN 0-7514-
0032-7.
[89] The Wind Power - Wind Turbines and Wind Farms Database.
http://www.thewindpower.net/index_en.php.
[90] A. F. Tol, van, Foundation Engineering and Underground Construction, 2006. Delft
University of Technology, Course CT5330.
[91] A. Verruijt, Soil Mechanics, 2001. Delft University of Technology, Course CT2090.
[92] , Offshore Soil Mechanics, 2006. Delft University of Technology, Course OE4624.
[94] , Strain Accumulation in Sand due to Drained Cyclic Loading: on the Effect
of Monotonic and Cyclic Preloading (Miner’s Rule), Soil Dynamics and Earthquake
Engineering, 30 (2010), pp. 736–745.
63
Appendix A
This Appendix contains borehole logs and CPT results of a North Sea location used for the
OWEZ wind farm. The mono-pile foundation used at this site is introduced in Chapter 1
and further analysed in Chapter 3.
65
66 Appendix A
This Appendix presents several plots in which the diameter D influence is plotted against
depth z. In Chapter 2 the parameters depending on the diameter D and depth z, A(z),
pus (z) and pud (z), which form the basis for p-y curves, are presented. Figures B.2 and B.3
show that at a depth greater than 12 m below soil surface there is no difference between
static loading and cyclic loading.
0
p , D = 4.4 m
us
pud, D = 4.4 m
−5 pus, D = 2.2 m
pud, D = 2.2 m
Depth [m]
pus, D = 1.1 m
−10
pud, D = 1.1 m
−15
−20
0 2 4 6 8 10 12
Soil resistance [kN/m] 4
x 10
Figure B.1 pus and pud as function of D plotted against depth (z)
77
78 Appendix B
−5
Depth [m]
−10
Astatic, D = 1.1 m
−15
Astatic, D = 2.2 m
Astatic, D = 4.4 m
−20 Acyclic
Figure B.2 Factor A to account for loading type plotted against depth (z)
4
x 10
2
s, 18 m
s, 11 m
1.5 s, 4 m
c, 18 m
c, 11 m
p [kN/m]
c, 4 m
1
0.5
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Deflection [m]
Figure B.3 Differences between static and cyclic p-y curves at different depth (z)
Based on the CPTs as shown in Appendix A an estimated soil profile qc is created. The
qc,1 , qc,2 and qc,3 refer to CPTs W-CPT08, W-CPT09 and W-CPT36 respectively. The
qc values is used for MPile calculations for North Sea (OWEZ) conditions. The results of
these calculations are the cap plots shown as Figures 3.1b and 3.1a.
0
qc,1
−4 qc,2
−8 qc,3
qc
−12
Depth [m]
−16
−20
−24
−28
−32
0 10 20 30 40
Cone resistance [MPa]
Figure C.1 Estimated CPT values for North Sea sand at the OWEZ wind farm
79
80 Appendix C
This Appendix presents an overview of MPile calculations performed to indicate the lateral
pult of 2.2 m and 4.4 m diameter mono-piles. In Chapter 3 different topics concerning a
laterally loaded large diameter mono-pile in sand are covered.
81
82 Appendix D
Figure D.1 Ultimate pile deflection, moments and shear forces of laterally loaded 4.4 m mono-
pile with a load excentricity e of 0.0 m
Figure D.2 Ultimate load cap of 4.4 m mono-pile with a load excentricity e of 0.0 m
Figure D.3 Ultimate pile deflection, moments and shear forces of laterally loaded 4.4 m mono-
pile with a load excentricity e of 2.4 m
Figure D.4 Ultimate load cap of 4.4 m mono-pile with a load excentricity e of 2.4 m
Figure D.5 Ultimate pile deflection, moments and shear forces of laterally loaded 4.4 m mono-
pile with a load excentricity e of 4.8 m
Figure D.6 Ultimate load cap of 4.4 m mono-pile with a load excentricity e of 4.8 m
Model piles
Figure E.1 gives details of the 2 model piles used for the ng experiments. In Chapter 3
the testing programme is introduced and Chapter 4 contains the test results.
89
90 Appendix E
14.00 28.00
13.00 26.00
35.00
35.00
0.50 1.00
68.40
0.34
136.80
0.68
13.68
27.36
Pycnometer results
The specific density of the sand particles that have been used for centrifuge testing is
examined using a QuantaChrome Ultrapycnometer 1000. Thereafter, the test series and
results are shown in Table F.1. In Chapter 3 additional soil tests are discussed. The
average density is determined using the mean of the last 3 measurements for both series
and equal to 2.6457 Mg/m3 . The (room) temperature at which the tests were performed
was 23.4◦ C.
91
92 Appendix F
This Appendix contains the steps performed analysing the field data using MatLab [88].
Results are presented in Chapter 6.
OWEZ data Two wind turbines of the Offshore Wind farm Egmond aan Zee have
been instrumented with accelerometers, anemometers, a thermometer and several devices
measuring nacelle and rotor related information. All recorded data is stored in a single file
for each 10-minute period.
The aim of this analysis is to derive pile displacements at mud line or seabed level
during strong wind or storm conditions. To achieve this, the 64 Hz North-South (NS) and
East-West (EW) acceleration data of the tower is used. Since the available data could
not be processed at once, each 10-minute periode file is analysed individually first, storing
the minimum, maximum and mean of each file in a new vector. This vector has been
examined and 20–24 November 2008 is chosen is reference period. The following steps
were performed to find the corresponding displacements.
- Open the consecutive 10-minute period files from 20–24 November 2008, extract the
acceleration information and store this data in a single vector. Accelerations in NS
and EW direction are treated separately.
- A periodic moving window technique with a 50% overlap (see Figure G.1) is applied
and the NS and EW signals are filtered using a low pass filter. The window size was
512, i.e. 8 seconds.
This periodic Hanning technique relies on analysing subsequent parts of the total
data file. A 50% overlap is used which results in an amplification factor of 1 for
the entire data range except both ends. By choosing a relatively small window size
compared to the total vector length (i.e. 22·106 ) the disturbed data is kept to a
minimum. 512 data points in this case make up a negligible percentage.
Although frequencies in excess of 10 Hz are not likely to be expected offshore or from
small magnitude, a small peak in the frequency spectrum was found around 23 Hz.
Signals >24 Hz are filtered out by a stable low pass filter.
93
94 Appendix G
1.2
Window 1 Window 2 Window 3 Sum
0.6
0.4
0.2
0
0 100 200 300 400
Data points [−]
Figure G.1 Principle of splitting and re-assembling data using consecutive 50% overlapping
windows
The average of the signal was positive, in the order of 0.08 m/s2 . Although a positive
mean of a signal does not make double integration impossible, the obtained displace-
ments would not make sense, i.e. displacements of several dozen metres are deemed
impossible by commom sense. Therefore, the mean of the signal was subtracted
before integration, which also is a common applied filtering technique.
Having applied a low pass filter and mean subtraction for each 512 long sub signal,
the entire signal is re-assembled.
- The re-assembed signal is integrated once to get velocities in m/s and a 5-point
smoothing is applied.
- The second integration step is performed and displacements in m are the result.
- A stable high pass filter is applied with a pass frequency of about 0.03 Hz.
- The displacement signal is analysed and the dominant frequencies are determined
using a fast fourier transform (FFT).
This Appendix contains results of performed triaxial tests including Mohr circles and
plots of ε versus the deviatoric stress (σ1 - σ3 ). Tests have been performed with different
conditions, although mainly with a cell pressure of 500 kPa. In Chapter 3 the soil properties
are discussed. The legends of Figures H.1 and H.2 refer to this Table H.1, which shows
test conditions.
In the following table the σ3′ is the effective cell pressure. The porosity and void ratio
are represented by n and e respectively. The h and w refer to the sample height and the
sample weight. Rate indicates the rate at which the deformation is applied. Finally, Avg.
and SD are used to represent the average and standard deviation respectively for several
relevant parameters.
95
96 Appendix H
1500
1000 c
d
σ1 − σ3 [kPa]
e
f
g
h
i
500
j
k
l
m
n
o
0
0 2 4 6 8 10 12 14 16 18 20
εzz [%]
1500
1000 c
d
σ − σ [kPa]
e
f
3
g
h
1
i
500 j
k
l
m
n
o
E50
0
0 1 2 3 4 5
εzz [%]
Figure H.2 Stress-strain curve with a cell pressure of 500 kPa, 0–5% strain
τ [kP a]
800
400
ϕ = 35◦
σ1 − σ3 [kP a]
0 400 800 1200 1600